PhilWIP

PCB Repair: Gradius III

Problem

No sound.

Diagnosis

Many components in the sound section had been replaced during a prior repair attempt:

G3_1 — Typical Konami sound system c. 1989: 007232, YM2151 and Z80

Acting on the assumption that these components were good and that the sound CPU (Z80) was working as expected, I instead focused on the communication mechanism between the main CPU and sound CPU.

Schematics for Gradius 3 are unavailable but the sound section is of a similar design to other Konami games of that era for which schematics are available (e.g. Final Round).

Data is passed to the sound CPU through an 8-bit register (74LS374 at F13). The main CPU writes a sound code into the register and then asserts the sound CPU’s /INT line. The sound CPU processes the interrupt request and reads the value from the sound register. Here are the relevant signals viewed using a logic analyzer:

gradius3_3 — Logic analyzer trace showing communication between main CPU and sound CPU

Here the main CPU writes 01h, which is latched into the register on the rising edge of ‘SOUND_LATCH_CLK’. The sound CPU interrupt line is asserted sometime afterwards:

gradius3_2 — Main CPU writes 01h into the sound register

Here is the sound CPU reading the register, during which ‘SOUND_LATCH_#OE’ is asserted. We expect to see 01h but instead the value is FFh:

Here is the suspect component, one of the few in the sound section which hadn’t been replaced. Ironically, it appeared to be the only non-Fujitsu 74LS IC on the entire PCB…